Testing the Effectiveness of a Small Group Substance Use Prevention Component Using Propensity Score Matching

Because the keepin' it REAL model prevention program takes place in regular classrooms, teachers are often faced with delivering curriculum to students with a wide array of substance use risk behaviors. Due to this, a small group component, REAL Groups, was developed to enhance the effectiveness of keepin' it REAL for the most at-risk students. A total of 109 7th grade students were referred to and participated in REAL Groups. The remaining 252 7th grade students who did not participate served as the basis for the control group. To account for biased selection into REAL Groups, propensity score matching (PSM) was employed. Once propensity scores were obtained, balance achieved, and common support attained, nearest neighbor matching without replacement was used to match students from the treatment group to the control group. Once matched, the final sample size was 108: 54 from the treatment group and 54 from the control group. The estimated average treatment effect on the treated for alcohol use at the end of 8th grade was calculated. Results indicated that alcohol use decreased among students who participated in the REAL Groups relative to matched students who did not participate (ATT=-.166; p=.05). By using PSM this study accounted for problems that arise in non-experimental designs with non-equivalent comparison groups. Using PSM allowed the researchers to create two groups, treatment and control, which were not statistically different from each other. This allowed for a less biased effect of participating in the REAL Groups, which showed that participation decreased alcohol use.